{import:NavigationMenu}
h2. Content
{toc:minLevel=3|maxLevel=3}
h3. Megasite Characterization
{html}
<span class="midtitle">Megasite characterization<br /><br /></span>
<strong>Topography</strong><br/><br/>
The map below shows the topography of the Rotterdam megasite.
{html}
\\
\\
!fig_21e_R_1_max.jpg|border=0!
\\
\\
{html}
<strong>Functions map</strong><br/><br/>
Industrial activities form the basis for defining the contaminant distribution. The following main activity types have been identified as being representative for the contaminant distribution:
<ul type="disc">
<li>Refinery</li>
<li>Storage of petroleum products</li>
<li>Storage of petroleum products and chemicals</li>
<li>Production of chemicals</li>
<li>Dry bulk and containers</li>
<li>Old industry (sites in Eastern harbours)</li>
<li>Other and unknown activities</li>
</ul>
This initial activity classification has been made for all major sites in the Rotterdam harbour, based on generic information from the Port of Rotterdam. The classification is under continuous revision due to developments in the Rotterdam harbour and changing activities at site level.<br/><br/>
<ul>
<li><a href="#" onCLick="show_window('../images/examples/rot/
<br>
Activities Maasvlakte
{html}
\\
\\
!fig_21e_R_15.jpg','rot',840,400)">Activities Maasvlakte</a></li>
<li><a href="#" onCLick="show_window('../images/examples/rot/_max.jpg|border=0!
\\
\\
{html}
Activities Europoort
{html}
\\
\\
!fig_21e_R_16.jpg','rot',840,400)">Activities Europoort</a></li>
<li><a href="#" onCLick="show_window('../images/examples/rot/_max.jpg|border=0!
\\
\\
{html}
Activities Botlek
{html}
\\
\\
!fig_21e_R_17.jpg','rot',840,400)">Activities Botlek</a></li>
<li><a href="#" onCLick="show_window('../images/examples/rot/_max.jpg|border=0!
\\
\\
{html}
Activities Pernis
{html}
\\
\\
!fig_21e_R_18.jpg','rot',840,400)">Activities Pernis</a></li>
<li><a href="#" onCLick="show_window('../images/examples/rot/_max.jpg|border=0!
\\
\\
{html}
Activities Eastern harbours
{html}
\\
\\
!fig_21e_R_19.jpg','rot',840,400)">Activities Eastern harbours</a></li>
</ul>
<br/><br/>_max.jpg|border=0!
\\
\\
{html}
<strong>Receptors</strong><br/><br/>
Receptors and their characteristics are are described in the <i>Starting IMS -> Building a conceptual model</i> and not further characterized for the Rotterdam megasite.
{html}
h3. Clustering
{html}
<span class="midtitle">Clustering<br /><br /></span>
<strong>Clusters in Port of Rotterdam</strong><br/><br/>
Clusters are distinguished to delineate zones within the megasite that have distinct characteristics such as: functional use, contaminant type and extent, geochemistry and redox conditions, or geohydrology. Because of these characteristics, the clusters will also have a typical risk profile which is related to the mass fluxes towards the receptors. These differences in risk are the most important factor for the definition of management options at cluster level. The suitability and effectiveness of management options varies between the different clusters. For example, source removal leads to a considerable risk reduction in a certain cluster, while in another zone plume treatment with enhanced natural attenuation appears to be more effective.<br/><br/>
In addition to risks, organisational aspects form an equally important criterion for clustering. These organisational aspects may include: similarity in the existing problems, the possibility to combine remedial measures and the willingness to cooperate with the neighbouring sites. The industrial companies will only commit to be part of a cluster if there is a benefit to tune the risk management scenarios with the other companies within a cluster. In other words there must be an increase of the cost-efficiency by a collective approach within a cluster.<br/><br/>
As basis for the delineation of clusters, the classification in figure presented below is used. <br/>
<br/>
<center>
<a href="#" onCLick="show_window('fig_22e_R_1.htm','rot',840,505)">
<img src="../images/examples/rot/fig_22e_R_1.jpg" border="0" alt="click on to enlarge the picture">
</a>
</center>
<br/>
<br/>
This classification has been derived by the Port of Rotterdam and reflects those parts of the harbour that are commonly recognised as functional units. In most cases they represent different sites that have the same type of industrial activity, historical use or period of establishment. As a consequence these zones reveal the different contaminant characteristics. Together with differences in geohydrology and geochemistry, this will determine the typical emission pattern towards the receptors. A further subdivision of these clusters is planned for the future, also taking into account organizational aspects.<br/><br/>
{html}
h3. Modelling
{html}
<span class="midtitle">Modelling<br /><br /></span>
<b>Modelling the transport process</b>
<br><br>
The objective is to make a prediction of the contaminant transport starting at the source of contamination towards one of the three defined planes of compliance (conceptual model). Two well-known modelling tools for calculating transport processes are MT3D and RT3D. Both use the groundwater fluxes as calculated by the groundwater model MODFLOW. Although MT3D and RT3D are widely applicable, both are used to simulate transport at site-scale only. In this particular case the groundwater model covers a whole region and can be characterised as a large-scale model. Because of the number of total model cells (~ 10,000) and the possible long simulation time (> 1000 years) it is not feasible to use the RT3D code for this problem. Therefore a simplified method is chosen: particle tracking in combination with pathline analysis. <br><br>
<b>Particle tracking and pathline analysis</b>
<br><br>
For the particle tracking part the existing code MODPATH<sup>7</sup> is used. It calculates the path of a particle through the flow domain using the fluxes as calculated with the groundwater model. Software, newly designed for this project, is used to analyse the pathlines and to calculate the process of biodegradation in time. A description of the modelling concept is given below.
<br><br>
Particles start in contamination zones that are based on spatially variable concentration grids. These grids are defined for the top of the Holocene layer and have a grid cell size of 150x150 m. In every grid cell the contaminant concentration is given and one particle is started for a concentration more than zero. Starting at these zones, the path of each particle through the model domain is calculated. Using this information the degradation of contaminant was calculated along the individual pathline using a first order irreversible degradation given by:
<br><br>
<center>
<I>C<sub>t</sub> = C<sub>0</sub> </I> x <I>e<sup>-(tk)</sup></I>
</center><br><br>
with:<br>
<I>C<sub>t</sub></I>: concentration at time t [mg/l]<br>
<I>C<sub>0</sub></I>: start concentration [mg/l]<br>
<I>t</I>: time [day]<br>
<I>k</I>: degradation rate [1/day]<br><br>
For every particle the starting concentration is known while it starts somewhere in the concentration grids. Using the calculated travelling time to the first point on the pathline the formula gives the concentration <I>C<sub>t</sub></I> on time <I>t</I>. This concentration (<I>C<sub>t</sub></I>) is the same as the new starting concentration (<I>C<sub>0</sub></I>) for calculating the concentration at the second point on the pathline. This procedure is repeated until the end of the pathline. A next point on a pathline is in general a crossing between two model cells. The value for the degradation rate (<I>k</I>) at every point in the domain is dependent on the local redox situation that is defined for the whole Rotterdam region. Given a redox situation in a model cell, ranging from aerobic to anaerobic, the matching degradation rate defined for every contaminant can be found and used for calculation. <br><br>
When a particle reaches a model cell that belongs to one of the planes of compliance the 'particle concentration' at that particular position is known. The mass flux (mg/day) as represented by the particle can be derived by considering both the particle concentration (mg/m<sup>3</sup>) and the representative water flux (m<sup>3</sup>/day). A first estimation of the representative water flux for a particle is given by the calculated vertical groundwater flux. The assumption is that a vertical flux is dominant within the top layers, which is confirmed by the flux pattern calculated with the groundwater model. <br><br>
Another calculation result is the concentration at one of the planes of compliance. More than one particle can reach the same model cell within such a plane. Therefore a total concentration C (ML<sup>-3</sup>) in a cell <I>j</I> is calculated based on the mass contribution of each incoming particle <I>I</I>:
<br><br>
<center>
<img src="../images/examples/rot/fig_23e_R_1.jpg">
</center>
<br><br>
This individual mass contribution <I>M<sub>I</sub></I> is derived from both the particle concentration <I>C<sub>I</sub></I> and the particle flux <I>Q<sub>I</sub></I>. <I>Q<sub>I</sub></I> represents the net water flux in cell <I>j</I>. Note that the concentration is calculated for one cell which means that the concentration in a vertical plane is the mean concentration for an area of at least 150x150 meter. The concentration in a cell within the vertical plane (first aquifer) is the integrated concentration for the complete layer thickness.
<br><br>
<b>Assumptions </b>
<br><br>
The application of this method implies some simplifications, which are explained in this paragraph.
<br><br>
<u>Dispersion and retardation</u><br>
The process of dispersion of the contaminant is not taken into account because the total transport is calculated for an area of 150x150 meter. On that (lumped) scale dispersion is not relevant. Retardation of contaminants is caused by sorption on the soil matrix, and calculated according to the formula below. Retardation depends on the type of contaminant and on the organic carbon content. It is assumed that the organic carbon content in the Holocene is 1% and in the Pleistocene aquifer 0.1%. These values are typical for these formations in The Netherlands and also expected for the Rotterdam region<sup>12</sup>.
<br><br>
<center>
<I>R</I> = <I>1</I> + <I>(rho/n)</I>*<I>Kd </I>
</center>
<br><br>
With; <br>
<I>Rho</I>: buldensity <br>
<I>Kd</I>: distribution coefficient <br>
<I>n</I>: porosity <br><br>
And: <br><br>
<center>
<I>Kd </I>=<I> foc</I> * <I>Koc </I>
</center>
<br><br>
With; <br>
<I>Koc</I>: organic carbon partitioning coefficient for compound of interest <br>
<I>foc</I>: fraction of organic carbon in the soil, <I>foc</I> = 0.59 * organic carbon <br><br>
<u>Sequential biodegradation</u><br>
During the biodegradation process daughter products can be formed for PCE that degrades to VC via the daughter products TCE end DCE. These daughter products have different degradation kinetics but within the modelling concepts the formation and degradation of these products are taken into account.
<br><br>
<u>Vertical drainage by sand piles</u><br>
The contaminant transport is influenced by the presence of sand piles. Different types of sand piles have been used for the compaction of newly retained land as well as for foundation purposes:
<ul>
<li>Short compaction drains</li>
<li>Long compaction drains</li>
<li>Foundation sand piles</li>
<li>Vertical rib drains</li>
</ul>
On a small scale, the influence of a sand pile on the transport of contaminant can be considerable. Within a zone around a sand pile the vertical ground water flow will pass the sand pile. In the anthropogenic layer, the water flows horizontally towards the sand pile, then it flows vertically through the sand pile and redistributes in the sandy layer below the sand pile. Outside that zone, the other part of the flux passes the Holocene layer. Because of the relatively high transmissivity of a sand pile the groundwater velocity are higher compared to the velocity through the Holocene layer. Because of these higher velocities, the travel times of contaminant that flows through the sand pile towards the first aquifer decrease, whereas the travel time of the contaminant that flows through the Holocene clay layer increases. Therefore the effect of sand piles has been taken into account within the transport modelling.<br><br>
<u>Pure phase contamination</u><br>
Pure phase contaminants act as a source for groundwater contamination. Therefore the presence of pure phase contaminants may become a risk especially when the contamination has a higher density than water (DNAPL's) and in combination with sand piles. A DNAPL can move through a sand pile in the direction of the aquifer by pure phase transport and form a new source for dissolved contaminant over a long time. Groundwater monitoring on several harbour locations proves that this process takes place.
<br><br>
However, the concept of pure phase transport is not built directly into the transport modelling. There is too much uncertainty about the distribution of this phenomenon over the harbour area because there is no detailed information about the exact locations of sand piles and the DNAPL. A DNAPL in the direct vicinity of a sand pile is more likely to show pure phase transport than through the Holocene where transport might be hindered by clay layers. To be able to assess the effect of pure phase transport on the contaminant situation, a pragmatic approach has been developed.
<br><br>
In the current transport modelling the contaminant sources in the anthropogenic layer are considered as black box and not taken into account. Instead, transport of contaminants dissolved in the groundwater is assumed to start from the top of the Holocene layer which means that particles for the particle tracking also start from there. In the case of a DNAPL, which is situated in an area where sand piles can be expected, pure phase transport is very likely and taking place rapidly. In these cases it is assumed that the DNAPL source is also directly present at the top of the first aquifer. As a first estimate, it is assumed that pure phase transport has taken place to the top of the first aquifer, as soon as the estimated concentration in the top of Holocene layer is larger than 1000 mg/l. It only takes place for contaminants that are heavier than water, such as: PCE.
<br><br>
<u>Start of contamination </u><br>
The Rotterdam Harbour has a long history, starting in the beginning of 1800 but with the most important extensions in the 20<sup>th</sup> century. This variability in time is built into the transport program. Four zones are defined with different periods of development (see figure below). However, for the modelling it is necessary to define just one representative start moment for every zone. The period of contamination probably started directly after the initiation. An early start of contamination results in an early breakthrough in the different planes of compliance, which is an overestimation. It is assumed that the contaminant load starts to build up in the beginning until a steady-state level has been reached. This moment has been considered as starting point. It was decided that the starting point for contamination is halfway between the establishment of the site and now. In the figure below the assumed representative start moment of the contamination is given corresponding with the period in which the bulk of the contamination was released.
<br><br>
<center>
<img width="100%" src="../images/examples/rot/fig_23e_R_2.jpg">
</center>
<br><br>
The modelling result comprise the calculated contaminant concentrations and mass fluxes at the three planes of compliance and can be presented for different moments in time. In this report the selected years are 1980, 2000, 2030, 2050, 2100 and <I>infinite</I>.
<br><br>
<b>Uncertainty analysis (Monte-Carlo)</b>
<br><br>
Uncertainty is essential for stakeholders to make decisions on risks and risk management scenarios. In earlier calculations uncertainty was taken into account by considering different scenarios: the best case, the worst case and the most likely case. Although results gave a range of outcomes, they did not give the probability of the outcomes. In order to obtain this information it is proposed to carry out a Monte Carlo analysis as part of the modelling.
<br><br>
In a Monte Carlo analysis the computer uses a random number generator to sample in an unbiased fashion values from the uncertainty distribution of a certain input parameter (e.g. contaminant concentration). These values will be used as input on grid cell level for the groundwater model. By repeating this exercise many times (e.g. 100 realisations), many different modelling outcomes will be obtained, which can be interpreted as an uncertainty distribution itself reflecting the combined uncertainty of all individual input parameters (see figure underneath).
<br><br>
<center>
<img width="100%" src="../images/examples/rot/fig_23e_R_3.jpg">
</center>
<br><br>
Previous modelling exercises have shown that the following input parameters cause the highest degree of uncertainty in the model, and therefore need to be considered in the Monte Carlo analysis:
<ul>
<li>Contaminanation (variation of concentrations) </li>
<li>NA conditions (variation redox conditions and biodegradation) </li>
<li>Geohydrology (effect of sand piles) </li>
</ul>
{html}
h3. Determination of Risk
{html}
<span class="midtitle">Determination of risk<br /><br /></span>
<strong>Determination of risk</strong><br/><br/>
Risk assessment can be done at several levels. The first step is the comparison of contaminant concentrations with risk-based screening levels. These levels are country specific, but are often expressed as target and intervention values. The values are based on the tolerable risks as determined for a worst-case scenario, which includes several different exposure routes for both human and ecological risk. In most countries the risk-based values are not function or land use specific. In the Netherlands, such values are only derived for immobile contaminants in the upper parts of the soil (Dutch: bodemgebruikswaarde). For the Netherlands an overview of risk-based values is given in table below.<br/><br/>
<table border="1" width="100%" cellspacing="0" cellpadding="2">
<tr bgcolor="#F4EBEA" align="center">
<td width="25%"> </td>
<td width="25%">Target Value (S)</td>
<td width="25%">In-between Value* (T)</td>
<td width="25%">Intervention Value (I)</td>
</tr>
<tr align="center">
<td>Benzene</td>
<td>0.2</td>
<td>15</td>
<td>30</td>
</tr>
<tr align="center">
<td>Naphthalene</td>
<td>0.01</td>
<td>35</td>
<td>70</td>
</tr>
<tr align="center">
<td>PCE</td>
<td>0.01</td>
<td>20</td>
<td>40</td>
</tr>
<tr align="center">
<td>VC</td>
<td>0.01</td>
<td>2.5</td>
<td>5</td>
</tr>
<tr align="center">
<td>1,2-DCA</td>
<td>7</td>
<td>203</td>
<td>400</td>
</tr>
<tr align="center">
<td>MCB</td>
<td>7</td>
<td>93</td>
<td>180</td>
</tr>
</table>
<sup>
* In-between value= (Target Value + Intervention Value) / 2</sup><br/><br/>
Stakeholders in the Rotterdam harbour area have discussed the use of these values as basis for defining risk-reduction objectives. It was concluded that for the Rotterdam harbour, in which the groundwater is strongly affected by industrial activities, the stringent target values are not realistic. Instead, the intervention value forms a much better criterion for the part of the deep groundwater directly below the harbours (2<sup>nd</sup> plane of compliance), and the deep groundwater outside the harbour area (3<sup>rd</sup> plane of compliance). The modeling results are represented graphically as the total impact of all priority contaminants on the 2<sup>nd</sup> and 3<sup>rd</sup> plane of compliance as a function of time. <br><br>
<b>Impact on top of the first aquifer (2<sup>nd</sup> plane of compliance)</b>
<br><br>
Modeling results indicate that the 2<sup>nd</sup> plane of compliance is impacted by contamination and that this impact increases in time (figure below). The increase rate is high between 1980 and 2030 and slows down after 2030. The impact is expressed as the percentage of the 2<sup>nd</sup> plane of compliance that has a concentration higher than intervention value. In total the area of the 2<sup>nd</sup> plane of compliance covers the total aquifer below the industrialized harbour, which is approximately 5000 ha.
<br><br>
The degree of uncertainty of the results is indicated by an uncertainty range. The impact on the 2<sup>nd</sup> plane of compliance at 2030 varies between 9.5% (25<sup>th</sup> percentile) and 12.5% (75<sup>th</sup> percentile). In the most likely situation (50<sup>th</sup> percentile) the impact in the year 2030 is 11.5%.
<br><br>
In the following order the priority contaminants contributed to the impact on the 2<sup>nd</sup> plane of compliance: benzene, naphthalene, xylenes, cis-DCE, VC, 1,2-DCA, ethylbenzene, 1,1,1-TCA, PCE, diCM, toluene, triCM and TCE. It can be observed that both petroleum hydrocarbons and chlorinated hydrocarbons contributed. In the case of benzene, naphthalene and xylenes the contribution can be explained by a combination of high concentrations in the anthropogenic layer and unfavorable biodegradation. The relative strong contribution of cis-DCE and VC is caused by the favorable biodegradation of PCE and TCE in which these contaminants are produced.
<br><br>
<table border="1" width="100%" cellspacing="0" cellpadding="2">
<tr bgcolor="#F4EBEA" align="left">
<td width="100%"> <b>Important notes:
<ul>
<li> The impact on the 2<sup>nd</sup> plane of compliance is based on the selected priority contaminants and does not take into account the contribution of site-specific "exotic" contaminants. </li>
<li> The impact on the 2<sup>nd</sup> plane of compliance is expressed as the concentration at the top of the aquifer as a function of time. It does not indicate the total mass that enters the whole aquifer.</li>
</ul></b></td>
</tr>
</table>
<br><br>
<img src="../images/examples/rot/fig_24e_R_3.jpg">
<i>Prediction of the impact at the 2<sup>nd</sup> plane as a function of time. The thick line indicates the most likely impact (at the 50<sup>th</sup> percentile), the thin lines the uncertainty range (respectively at the 25<sup>th</sup> and 75<sup>th</sup> percentile).</i>
<br><br>
In order to get an impression of the spatial distribution of the contaminant situation, the chance of exceeding the intervention value has been shown in the next figure. The contaminant situation has been predicted at different times. The figure below gives the chance of exceeding the intervention values of all priority contaminants and indicates the situation for the year 2030. The highest chances of exceeding the intervention values at the 2<sup>nd</sup> plane of compliance are present in the Eastern harbours, the Pernis area and the Eastern parts of the Botlek area. The Western parts of the Rotterdam harbour area have a lower chance of exceeding intervention values. This distinction can be explained by differences in NA and contaminant situation in the superficial layers. In general, the superficial contamination is stronger in the Eastern parts and the biodegradation of the most predominant contaminants (e.g. benzene) less favorable due to adverse redox conditions.
<br><br>
<a href="#" onCLick="show_window('../images/examples/rot/fig_24e_R_4_max.jpg','rot',840,400)">
<img src="../images/examples/rot/fig_24e_R_4.jpg" border="0" alt="click on to enlarge the picture">
</a>
<i>Spatial distribution of the chance of exceeding the intervention value in the year 2030.</i>
<br><br>
<b>Impact on the pristine groundwater systems outside the harbour area (3<sup>rd</sup> plane of compliance)</b>
The impact on the 3<sup>rd</sup> plane of compliance is predicted in time and expressed as the percentage of the boundary of the Rotterdam harbour that is affected by the sum of the contaminations above the intervention value (figure below).
Modeling results show that the impact increases in time from 2% in 2000 to 10% of the boundary in 2050 for the most likely situation (50<sup>th</sup> percentile). The uncertainty ranges are between 7% and 11% in 2050. Also after 2050 the impact continues to increase, indicating that due to long traveling times towards the 3<sup>rd</sup> plane of compliance not all contaminant sources have yet broken through. <br><br>
For the year 2030 the contaminant situation at the 3<sup>rd</sup> plane of compliance is shown in the final figure. In general the impacted length along the 3<sup>rd</sup> plane of compliance covers major parts of the harbor, especially the Pernis and Botlek area. In time the situation changes in such a way that both the total length along the 3<sup>rd</sup> plane of compliance and the chance of exceeding the intervention value increase. Compared to the 2<sup>nd</sup> plane of compliance, the chance of exceeding the intervention value at the 3<sup>rd</sup> plane of compliance is relatively high. In some cases the chance is higher than 50%. These high chances can be explained by the convergence of different groundwater flow paths at the 3<sup>rd</sup> plane of compliance, of which each contributes to the cumulative chance of exceeding the intervention value.
<br><br>
<img src="../images/examples/rot/fig_24e_R_5.jpg"><br>
<i>Prediction of the impact at the 3<sup>rd</sup> plane of compliance as a function of time. The thick line indicates the most likely impact (at the 50<sup>th</sup> percentile), the thin lines the uncertainty range (respectively at the 25<sup>th</sup> and 75<sup>th</sup> percentile).</i>
<br/><br/>
<a href="#" onCLick="show_window('../images/examples/rot/fig_24e_R_6_max.jpg','rot',840,400)">
<img src="../images/examples/rot/fig_24e_R_6.jpg" border="0" alt="click on to enlarge the picture">
</a>
<i>Chance of exceeding the intervention value at the 3<sup>rd</sup> plane of compliance in the year 2030</i>
{html}
h3. Finalize Clustering
{html}
<span class="midtitle">Finalize clustering<br /><br /></span>
A further refinement of the clustering is not considered as meaningful at this stage. Results of the risk assessment do not affect the earlier made cluster classification. At a later stage, a refinement of the clustering is foreseen when organizational aspects concerning the implementation of management scenarios become relevant.
{html}
|